01
Our FDEs talk with the team, trace how the business runs, surface unit economics, and connect the systems. Then we build the agents and automations that move workflow lift and EBITDA. Pricing is transparent at $750/hr, and we co-invest.
// truesilver partners
Enhance.
Automate.
Compound.
TS-0001 / Healthcare funds / Forward deployed engineers
We work with healthcare-focused funds and private equity firms. We turn AI and automation into measurable portfolio value. We start with the fund thesis, then forward deployed engineers (FDEs) embed inside real healthcare operations, connect the systems, and ship reviewed automations. Less administrative drag. More EBITDA.
01 / Thesis
AI in production. Workflow proof for fund level strategy.
02
We often deploy our own technology: STUDIO and STUDIO FOR FUNDS, so strategy moves from diligence into reviewed workflows.
03
Patient access, provider admin, scheduling, revenue cycle, intake, referrals, data warehouses, and data cleanup are the first workflows we automate.
04
We build with the full model stack, from frontier APIs to local models on practice hardware. Every agent ships with an owner, a review path, and a measured lift.
OpenAI
Anthropic
OpenRouter
Google
xAI
Meta
Mistral AI
DeepSeek
Moonshot AI
Alibaba Qwen
MiniMax
Cohere
Groq
Fireworks AI
Perplexity
NVIDIA
Hugging Face
Liquid AI
turbopuffer
DuckDB
Vercel
05
Clinical and operator judgment stays with people. We automate the administrative drag around it, and the owner can inspect every change.
02 / Engagements
Engagements
01
Portfolio engagement
Our forward deployed engineers (FDEs) embed in one portfolio company, connect the EHR, revenue cycle, scheduling, claims, CRM, and files, and ship automations a workflow owner reviews. The signal is time saved and EBITDA movement.
See pricing02
Sell-side accelerator
For owners and bankers 6 to 18 months before a sale. We reconcile the source systems into one warehouse, ship reviewed automations, and write the technology and data room section. Buyers inspect source, owner, version, review log, and code.
See the accelerator03
Diligence and quality of earnings
We review healthcare data, operations, technology, and workflow risk, and trace revenue, claims, and addbacks to source material. Fund teams build IC memos, LP decks, and portfolio notes in Studio for funds, with source trails and approval logs attached.
See Studio for fundsWhat we build
Read. Automate. Show the data.
01 / Read
The Read
We embed one forward deployed engineer (FDE) in a single company, trace the EHR, RCM, scheduling, and claims, and name the few workflows worth automating with the owner who would review each.
Start with the Read02 / Automate
Reviewed automations
We connect the EHR, RCM, scheduling, claims, and CRM, then ship focused agents a named workflow owner reviews before anything changes. Every output carries its source, owner, and review log.
See how we ship03 / Studio
The surgeon board
We build the board in Studio. An operator asks for RVUs by surgeon over the last 3 months, the LLM drafts it from the EHR and RCM tables, and the owner checks the numbers against source.
See Studio03 / Precise Moves
3%
Small moves matter when they change behavior inside a real workflow. Keep the durable core intact, alter the route, handoff, or review step, and measure whether the company actually runs better.
Method note
A fund strategy gets sharper when the intervention is small enough to inspect and built to compound.
Small moves for funds
For healthcare-focused funds, the method means choosing the few moves where AI, focused agents, technology, or automation can change a portfolio company workflow without disturbing the durable core: better intake, cleaner data, faster routing, tighter handoffs, or less administrative drag around care.
Patient access
Provider admin
Scheduling
Revenue cycle
Intake / referrals
Data warehouses
Data cleanup
04 / Criteria
Built for healthcare-focused funds with real portfolio work to do.
05 / Essays
Essays
01
Medical LLM Benchmark
Judging medical language models against the work they will actually touch.
Read essay
02
The Checklist Came Before the Model
Reliability on top of an unreliable component is treated as AI's newest engineering problem. Anesthesia solved it forty years ago and wrote down what worked.
Read essay
03
Run Controls Before Patient Samples
Every clinical lab proves its analyzer still tells the truth each morning before the first patient sample. AI engineering calls that eval-driven development and considers it advanced.
Read essay
06 / Dom Phillips
Dominic Phillips
Founder and Managing Partner
Dom was CTO at TeleClinic through its acquisition. After TeleClinic, he joined CompuGroup Medical, a German healthcare-software group with over EUR 1.2B in revenue, where he built the CGM One Phone Assistant and worked on AI strategy.
He co-founded CodeSubmit with his wife Tracy. Hiring teams at Apple, Netflix, the US Air Force, Genentech, and Pfizer used it. He also co-founded Mayze, a dating app, and was CTO at Revel, a fashion app for discovering new designers in Los Angeles. Both failed, and both were useful: Mayze taught him about matching algorithms, and Revel taught him that branding shapes how customers understand a product before they try it. It made him more deliberate about naming, positioning, and the first screen a customer sees.
Dom lives a quiet, productive, distraction-free life in the desert in La Quinta, California (fastest Internet and lowest electricity cost in CA). He enjoys spending time with Tracy and their son Tom, tennis, and video games.
Builder reference
“You are a real builder, not a deck-only guy. Strong technical spine, real healthcare operator and investor context, and you keep bringing AI back to measurable workflow and EBITDA impact. I would also say you are unusually comfortable inside messy real-world systems, which is rare and useful.”